Summary Electroencephalography (EEG) offers a promising modality for biometric identification, though balancing performance, interpretability, and robustness remains a challenge. High-accuracy methods typically rely on supervised or deep learning models with limited neurophysiological interpretability, while more interpretable feature-based techniques can be sensitive to data length and variability. This study presents a parameter-free framework using synchronization modes across multiple frequency bands. Following rigorous preprocessing, synchronization patterns were derived from five bands under eyes-open and eyes-closed conditions. Results indicate that synchronization modes converge to stable, person-specific patterns as EEG duration increases. Notably, beta- and gamma-band modes demonstrate strong stability, achieving 100% correct recognition rate for identification and 0% equal error rate for authentication on a publicly available dataset of 109 subjects. Lower-frequency bands exhibit slower convergence and reduced discriminability. The results highlight high-frequency synchronization modes as interpretable, stable, and highly effective EEG fingerprints, offering a robust alternative to learning-based biometric approaches.
Vaezi et al. (Mon,) studied this question.